The structure and composition of a wooded urban ravine system

Abstract

Fifty wooded ravine-slope stands in the Highland Creek valley system, Scarborough, Ont., were sampled for tree, sapling, shrub, and herbaceous vegetation by using plotless methods. In addition some soil characteristics were measured or estimated. Woody species were ordinated as stem size classes, using detrended correspondence analysis (DCA). Results indicated a strong trend along the first axis from pioneer species, typical of disturbed sites, through to shade-tolerant species typical of undisturbed forests. This pattern suggests that secondary succession is actively occurring in the woodland complex. However, little or no evidence of a self-regenerating "climax" woodland was apparent. Understorey vegetation, comprising 184 species in 48 of the stands, was also ordinated by DCA. These same data were subject to two-way indicator species analysis (TWINSPAN), a divisive clustering procedure. The ordination showed a distinctive separation along the first axis of wide stands from narrow stands with at least one edge exposed to full sunlight. The first division by TWINSPAN also segregated wide from narrow stands, listing those species which could be used as indicators of the degraded woodland condition. This results through stand isolation from neighbouring woodland and (or) increased lateral illumination. Furthermore, TWINSPAN indicated those woodland species under threat of local extinction in narrow, isolated stands. Compositional variation along the second DCA axis maximally discriminated among stands of the narrow, isolated group, although a clear underlying gradient was not apparent. A poor but detectable relationship emerged with an index of trampling. Canonical correlation analysis between the understorey vegetation ordination axes and 10 soil variables showed only the first canonical variate to be significant (r = 0.69, p = 0.0199). Major contributors to this were pH and the index of trampling intensity in the soils data. Because soil pH covaries with the survey design, variation in the understorey composition is attributed primarily to fragmentation and the sensitivity of woodland to urban pressure. Stand diversity versus quality as well as the sizes and shapes of urban woodlots are briefly considered.